Regression analyses Panel A

In Tables 6 and 7, the regression results for the full sample, including both large and SME firms, have been depicted. Here, it has been chosen not to include the industry dummies due to the high multicollinearity that has been caused by inclusion of these variables. Overall, the models differ in their level of R Squared, or explained variance. The best explanatory model is the one including the ASCL-index for the interest coverage ratio (i.e. R Square of 14,1%) and for the Altman’s Z- score, the best model includes DIO, DSO, and DPO with an explained variance of 25,7 percent. As can also be seen is that, contrary to what literature indicated, models including industry variables do not consistently perform better than models excluding industry variables in terms of explained variance.

42

Cash conversion cycle

It has been hypothesized that the cash conversion cycle follows a quadratic relationship with firm performance, as it has been argued that there is an optimal level of working capital (Baños- Caballero et al., 2014). In this study, the authors have found a positive relationship between CCC and firm performance, and a negative relationship between CCC2 and firm performance, indicating the existence of an inflection point at −𝛽1_/2𝛽2_{. However, in this sample it has been found that} there is a negative relationship between both CCC and Altman’s Z-score (t = -6,876, P < 0.01) and

43

between CCC2 and interest coverage ratio (t = -1,730, P < 0.10). When including CCC and CCC2 in the full regression model with Altman’s Z-score begin the dependent variable, both CCC (t = - 9,825, P < 0.01) and CCC2 _{(t = 2,061, P < 0.05) are statistically significant. When interest coverage} ratio is the dependent variable in the full regression model, only CCC is statistically significant (t

= -2,013, P < 0.05).

The results found in the regression results including the interest coverage ratio are partially as expected: the cash conversion cycle follows a quadratic relationship with the probability of bankruptcy in the individual model. The same holds true for Altman’s Z-score regarding the full

44

regression model. This is probably due to the costs of not investing the locked-up money elsewhere and is in line with results found in previous studies, e.g. Baños-Caballero et al. (2014) and Zeidan and Shapir (2017). In addition, it has been chosen to plot the variable CCC and the interest coverage ratio independently (not reported) and look at the F-statistics of both a plotted linear relationship and a quadratic relationship. Although both have a very low level of R Squared, i.e. respectively 0.000 and 0.003, the plotted quadratic relationship is statistically significant (P < 0.01) and yields a higher F-statistic, respectively 9.192 and 1.048. This is in line with the results found in the regression analyses with the dependent variable being interest coverage ratio. Applying the computation for the inflection point, i.e. 𝛽1/2𝛽2 shows that the optimal cash conversion cycle yields 82,97 days. This provides support for Hypothesis 1.

When plotting the relationship between CCC and Altman’s Z-score, different results as compared to the interest coverage ratio have been found as both the linear and the quadratic relationship have an R Squared of 0,009, which is significant (P < 0.01). However, the F-statistic of the linear relationship is higher than the F-statistic of the quadratic relationship, respectively 75,532 and 37,807, thereby providing evidence for there being a linear relationship between CCC and Altman’s Z-score. This would indicate that the relationship between the cash conversion cycle and Altman’s Z-score is negative, and that the longer it takes to collect cash, the higher the probability of financial distress. This has also been found by Eljelly (2004) and García-Teruel and Martínez-Solano (2007). It is difficult to find conclusive support for Hypothesis 1, as there is only partial support. However, what can definitely be taken from these analyses is that having a high level of CCC, and thus a high level of working capital leads to firms being more susceptible for financial distress.

Regarding the components of the cash conversion cycle, both analyses show that days’ inventory outstanding, days’ sales outstanding and days’ payables outstanding all have a statistically significant relationship with the probability of financial distress. Generally, it can be concluded that high numbers for all of these indicators means a higher probability of financial distress, as all are highly negatively related to both the interest coverage ratio and Altman’s Z- score, except for DSO and the interest coverage ratio (t = 2,298, P < 0.05). This may indicate that having many days’ sales outstanding is positively related to business performance as it shows that the company has a high revenue and just still needs to collect the money from it. However, overall it indicates that having a high level of receivables and inventory has a negative influence on firm

45

performance. In addition, having many payables outstanding follows the same relationship, as it may indicate that a firm has liquidity problems, which may lead to them having less access to financing or problems in general.

ASCL-index

The age-size-cash-flow-leverage index indicates the level of access to financing, of which a 4 indicates a high access to financing (i.e. above-median age, size, cash-flow and below-median leverage), whereas a 0 indicates a low access to financing (i.e. below-median age, size, cash-flow and above-median leverage). The ASCL-index has been found to relate positively with the interest coverage ratio (t = 19,531, P < 0.01) and Altman’s Z-score (t = 12,339, P < 0.01). This indeed indicates that firms having a better access to financing have a lower probability of financial distress. This might be due to the pecking order theory, which describes that if a firm does not generate enough internal funds, it has to reside to outside capital, preferably debt. If a firm is not able to collect outside financing, it will have a negative impact on the business, as a firm is then not able to expand or may run into liquidity problems. Therefore, the relationship between ASCL and financial distress behaves as expected in Hypothesis 2.

Interaction CCC and ASCL

In addition, we are interested in finding evidence for access to financing moderating the relationship between CCC and the level of financial distress. This has been included on the basis of agency problems, which increases the wedge between the costs of internal and external financing due to credit rationing (Baños-Caballero et al., 2014). It was argued that a higher level of working capital requires more financing, and thus it is argued that firms having lower levels of access to financing will have a lower level of working capital, and thus a lower cash conversion cycle. When including only the interaction variable CCC*ASCL, of which both variables were firstly centered, and the control variables, both individual regressions do not yield statistically significant results, but in the full models there has been found evidence of the interaction variable being statistically significant related to the interest coverage ratio (t = 2,022, P < 0.05) and Altman’s Z-score (t = 2,864, P < 0.01). This provides evidence for Hypothesis 3 and indicates that access to external financing moderates the relationship between the cash conversion cycle and the probability of bankruptcy.

46

When testing the interaction variable by looking at the F-change in both models, the results indicate that there is evidence for an interaction effect with the interest coverage ratio (R Square change = ,003, Sig. F change = ,043) and Altman’s Z-score (R Square change = ,003, Sig. F change

= ,004). This provides additional evidence for Hypothesis 3.

Industry barriers

Operating in an industry with many risks, will lead to a higher probability of default according to industry organization economics theory (Miloud et al., 2012). Therefore, a variable has been tested that identifies that overall percentage of barriers perceived in a certain industry in a specific province. The kind of barriers included in this research are demand risk, labour market risk, materials risk, financial risk and weather risk. The overall barrier-variable has a statistically significant negative relationship with the interest coverage ratio (t = -5,159, P < 0.01). The relationship between the percentage of barriers and Altman’s Z-score is insignificant (t = ,751, P

> 0.1). This provides some evidence for industry organization economics theory and therefore it can be concluded that Hypothesis 4 is somewhat supported.

When taking a closer look at the types of risks an industry is subjected to, it becomes clear that only financial risk and weather risk are significantly related to the dependent variables. The level of financial risk, and more specifically firm access to financing, is positively related to the probability of default as measured by interest coverage ratio (t = -3,651, P < 0.01). This has also been found by the variable “ASCL” which indicated that having a lower access to financing will have a positive influence on the probability of default. If a firm cannot obtain bank financing when internal funds do not suffice, a firm cannot grow or even be unable to pay its obligations, which will have a negative influence on firm performance. Furthermore, weather influences are statistically significant and positively related to the interest coverage ratio (t = -3,235, P < 0.01) and significantly positively related to Altman’s Z-score (t = 3,552, P < 0.01), which seems counterintuitive. A possible explanation for the latter result found is that some firms are better able to deal with weather circumstances than others due to the resource-based view (Peteraf, 1993).

Industry growth

The literature and empirical evidence have not found conclusive evidence of the relationship between the industry growth rate and the level of financial distress a firm encounters. It has been

47

argued that operating in a high-growth industry leads to firms striving for innovations (McDermott, 2014) and higher demand (Karuna, 2007) due to reduced further entry into the industry. However, such an industry will also be attractive for new entrants and may lead to a loss of market share and a higher price competition (Karuna, 2007). This inconclusive evidence is also reflected in the regression results. The relationships between industry growth and the interest coverage ratio (t = ,145, P > 0.1) and Altman’s Z-score (t = 1,300, P > 0.1) are insignificant. Therefore, no evidence has been found for Hypothesis 5, and it still remains the question whether industry growth influences the level of financial distress and if so, in what direction.

Industry competition

Regarding the level of industry competition and its influence on firm’s probability of default, the literature and empirical evidence were also inconclusive. On the one hand, it was argued that a high degree of competition should stimulate managers to perform better and thus a high degree of competition can have a positive influence on firm performance (Schmidt, 1997; Karuna, 2007). However, according to Sutton (1990), intense competition could also lead to bankruptcy as firms cannot compete efficiently. As the conventional view and literature supported that competition leads to a higher probability of default due to lower price margins, this has also been hypothesized. The relationship between competition and the interest coverage ratio is insignificant (t = -,464, P

> 0.1), but the relationship between competition and Altman’s Z-score is significant (t = 3,163, P

< 0.01). This provides evidence for Hypothesis 6, as firms that operate in an industry where the competition gets less, or where their own competitive position within the industry gets better, have a lower probability of default. This probably means that firms are able to attain market share and charge higher prices, so they do not have to compete on lower prices and become more efficient (Karuna, 2007). This is also in line with the univariate statistics performed as the correlation between a better position in the market and a higher sales price is significant at the 1%-level with a correlation of ,233.

Industry sales price

The sales price is linked to the level of competition present in the industry (Karuna, 2007). The prevailing argument is that higher levels of competition lead to lower sales prices and thus lower margins for firms. Therefore, it has been hypothesized that having lower industry prices leads to a

48

higher probability of default. However, the opposite has been found regarding the relationship between industry sales price and the interest coverage ratio (t = -2,101, P < 0.05). The relationship between industry sales price and Altman’s Z-score has been found to be insignificant (t = -,013, P

> 0.1). These findings indicate that a lower industry sales price leads to a lower probability of default, thereby not supporting Hypothesis 7. This might be due to firms operating in industries with decreasing sales prices having more efficient operations and are able to compete on price (Sutton, 1990).

Control variables

In the baseline model, it can be seen that the control variables all yield statistically significant relationships with both the interest coverage ratio and Altman’s Z-score. The variable “Age” is significantly and positively related to both the interest coverage ratio (t = 5,668, P < 0.01) and Altman’s Z-score (t = 2,790, P < 0.01), which was also expected. Older companies have oftentimes easier access to capital and have more knowledge built up, so that agency problems are oftentimes alleviated, which enhances firm performance (Mulier et al., 2016). The variable “Growth” follows a significantly negative relationship with the interest coverage ratio (t = -11,692, P < 0.01) and with Altman’s Z-score (-19,954, P < 0.01). This may be due to companies not having the necessary knowledge in-house to support the growth, which in turn may harm firm performance and cause financial distress (Marcelino-Sádaba et al., 2014).